Embed Size (px)
Citation preview
© Copyright 2016 OSIsoft, LLCUSERS CONFERENCE 2016
© Copyright 2016 OSIsoft, LLCUSERS CONFERENCE 2016
Presented by Kyle Gogolinski
Peter Wright
Mining Equipment Event Synthesis
Early Intervention for Increased Efficiency
Mining Equipment Event Synthesis
Early Intervention for Increased Efficiency
Kyle GogolinskiProcess Control & Automation
Peter WrightDexcent
Presentation for OSIsoft UC2016
Conference
April, 2016
Background• Syncrude is a crude oil producer from the Athabasca
Oil Sands deposit in northeast Alberta. We operate
two large-scale surface mines (known as Mildred
Lake and Aurora) 35 km apart.
• Syncrude uses a truck and shovel technique for
bench mining of the oil sand deposit.
• Mining equipment uptime is a key factor in operating
efficiency, driving both production cost and
movement volumes.
• Optimized preventive maintenance programs and
just-in-time intervention are key to minimizing major
component failures
requiring days or weeks to repair.
• A vast knowledgebase has been created at Syncrude since
our inception of truck and shovel mining in the mid-1990s.
The Challenge
Effectively leverage our reliability
knowledge base by:
Transforming reactive, time-
intensive forensic data reviews
into automated, near real-time
Event synthesis and creation
to enable the next level of
mining equipment efficiency
in our harsh operating
environment.
Environmental Factors Impacting Asset Maintenance
• 136 unit heavy hauler fleet, plus shovels, graders, and dozers
• Climate extremes: ‒30°C (‒22°F) to +35°C (95°F)
• Geographic location and haul distances
• Oil sand ore
• Rock hard when cold
• Sticky when warm
• Always abrasive
The Challenge
• Effort-intensive manual analysis – truck sensor dataset
too cumbersome for timely analysis and intervention –
used for forensic analysis only.
• Limited subject matter expertise – attrition and loss of
knowledge a risk.
• Data stream and connectivity challenges result in
dropped, skipped, and prematurely terminated
calculations.
• Ability to integrate into existing workflows in an actionable
manner.
Business & Big Data Factors Impacting Asset Maintenance
The Challenge
The Team
Syncrude Canada Ltd.
• Process Control & Automation
• Equipment & Reliability Engineering
• Research & Development
Dexcent
• Architect
• Senior Developers
OSIsoft
• Sales
• Centre of Excellence
• San Leandro Engineers
The Plan
Overall Objective
• Create a Mobile Equipment Event Synthesis (MEES) solution to offer support to
the tracking, analysis, and reporting of mechanical events that occur on
Syncrude’s mobile equipment.
Pilot / Test Platform Phase Objective
• Implement extremely complex calculations with the intent to overload the
system. Demonstrate the ability for PI ACE to handle large volume data
calculations, inconstant data acquisition, and a large number of contexts.
Production Phase Objective
• Optimize for production use, streamline calculations, integrate with notification
systems, validate, and tune performance.
The Requirements
• Control the execution of calculations to only process when the data is
there. Need to ensure that no gaps in event computation are created as a
result of trucks moving in and out of coverage area.
• Process large volumes of data as it comes into the system. The average
processing volume during the pilot was 1716 values/second. Surge
volumes were much higher – a single unit offline for an hour results in a
79K data value surge when it comes back online.
• Create standalone components for each use case to support
independent evaluation, schedule tuning, and overall solution
maintenance.
• Produce actionable output.
The Approach - Key Concerns
• Handling of incoming data would be needed to ensure that no
gaps in event computation were created as a result of trucks
moving in and out of coverage areas.
• The ability to process large volumes
of data as it came into the system would be required for
quick event identification and timely intervention.
• The delivery of trustworthy, actionable results would be needed for
acceptance.
The Approach
• Analyze what has worked and what has not.
• Work closely with E&RE to create extremely difficult test use case
patterns.
• Syncrude and Dexcent would work closely with the OSIsoft Centre of
Excellence team, as well as the San Leandro Engineering team, to solve
the data gap challenge.
• Implement the use cases, test the pilot, evaluate the results.
• Revisit the use cases and modify the calculations for efficiency in the
production environment.
• Validate the production solution.
• Transition to the PC&A support team.
Event Synthesis – Pilot Use CasesSub-system Use Cases Condition/Action
General Monitoring Out of Wireless Range Truck out of range and/or wireless system disconnected
Acquisition Offline Validate data and ensure system readiness
Engine Throttle Position Condition Throttle position minimum threshold under load
Turbo Failure Turbo boost pressures under load
Injector Failure Engine injector failure
Engine Oil Dilution Ensure fuel or coolant is not diluting engine oil
Coolant Temperature Delta Variations in coolant temperature between the engines
Power Train Torque Convertor Overheat Torque converter temperature limits and location
Braking Service Brake Applied at Speed Service brake applied while in motion
Brake Overheat Brake temperature
Braking Pumps Cycle Brake accumulator pressure cycle faults
Frame Improper Strut Charge Strut is properly charged
Strut Deflation Strut has lost all pressure
Airborne Front or back of the unit is airborne
Side Load Payload was loaded off to one side
Front Load Payload was loaded too far forward
Structural Force Structural force (pitch, rack, roll) exterted on the frame
Abusive Dumping Dumping procedures were followed
Abusive Loading Loading procedures were followed
Steering Steering Pumps Cycle Steering pump short cycles and pumps constantly on
Event Synthesis – Pilot Results
• All use cases were implemented successfully.
• PI ACE processing load, scaled across two schedulers, was adequate to process
regular data flow.
• When load balanced, the average CPU utilization was ~25% for data flow from ~80
trucks.
• When processing backlogs, schedulers used 100% CPU.
• Data from 44 tags per truck was used to calculate the use cases.
• Events raised were inserted into a database table.
• Single instance alarms were produced.
• Delivered condition reminder notifications for long-running events.
• All risks and concerns were successfully addressed in the pilot
phase.
• The creation of the production version would focus on
streamlining the calculations;
refactoring the code; extensive
testing; and performance tuning.
• A shortened use case list would be
adopted.
Event Synthesis – Transition
Event Synthesis – Production Goals
Where the Pilot phase was intended to prove the system
capability, the Production phase focused on:
• Efficiency - reduced execution times; increased performance.
• Integrity - business value would only be realized as confidence
and trust increases.
• Effectiveness - implementing additional notifications and
escalations; adding cumulative event handling; and integrating
with the Maintenance systems and processes.
Event Synthesis – Standards
The core guidelines and standards for the production release
required that:
• Business validation would be performed on all use cases for
correct logic implementation.
• Continuous monitoring and QA checks would be done to
raise confidence in the results produced.
• Event data would be integrated into existing visualization
platforms and operator care workflow before full adoption
would be realized.
Event Synthesis – Production
Use Cases Condition/Action Assets Monitored
MDSP Offline Monitor data buffering, connectivity, and data age to automate
wireless system health .
150
Lubrication Cycle Time Monitor lube injection frequency to ensure proper lubrication
injection rates.
85
Engine Injector Failure Monitor the asset for possible engine injector failures 93
Engine Oil Dilution Monitor engines to ensure fuilds are not leaking into the engine
lubrication system and diluting the engine oil.
132
Steering Pump Cycle Steering pump pressure cycle analysis to determine the health
of the steering pump.
79
Service Brake Pump Cycle Brake accumulator pressure cycle analysis to determine health
of brake charge.
79
Suspension Charge Monitoring Monitor an assets suspension to ensure the struts are properly
charged.
119
Dumping Procedure Monitor that the operator dumping procedure is followed,
preventing injury from jarring or loafing.
52
The Results - Quantification
The effectiveness of change in maintenance and reliability programs
can be a challenge to capture and validate.
• Rather than focusing on the savings realized per use case or
event, thanks to years of data collection, we were able to look at
overall operating cost reductions.
• There had been no other significant change in maintenance cost
reduction programs in the past year, therefore any major cost
reductions could be attributed to the MEES project.
The Results – Production
• Over 6600 data points collected and analyzed from 131 heavy
haul trucks and 5 shovels.
• Critical event escalations implemented.
• Cumulative events monitored and reported.
• Notifications, Alerts, and Alarms displayed in monitoring platform.
• Daily summary report on wireless status automatically delivered.
• Dumping compliance reports emailed.
• Calculated fleet operating expense savings of $16.75/hr/unit
equating to a $20 million annual operating cost avoidance.
• Reduced non-procedural dumping incidents by 85% in the first
month in production.
Operator Control Panel
Control Panel
• Asset Health
• Log access
• Event access
• Notifications
• Report access
Operator Control Panel
Detail drill-down
Monitoring the Monitoring
PI ProcessBook
• Calculations
• Performance
• Pointsource
• Statistics
The Results – An ExampleFailure: Heavy Hauler Engine Ventilation Event
Post pilot and prior to the production implementation of MEES:
• A unit experienced a catastrophic engine failure where “the internal parts of the
engine disconnected from their normal position and made their way outside,
leaving a large “ventilation window” in the side of the engine block”.
• Manual forensic analysis determined a valve had broken, severely damaging a
cylinder. The valve then passed through the exhaust port and damaged a turbo
charger. The damage to the cylinder led to the ventilation event. The cause of
the valve failure: lack of lubrication.
Running MEES calculations over the event timeframe revealed:
• An Engine Injector Failure event would have been triggered well in advance.
• An Engine Oil Dilution alarm would have been raised 2.5 days before the
engine failed.
Ventilated Engine Example
Syncrude was able to effectively
leverage our reliability knowledgebase
by transforming reactive, time-intensive
forensic data reviews into automated,
near-real-time event synthesis and
creation.
This enabled the organization
to move the next level of
mining equipment efficiency
in our harsh operating
environment.
Mission Accomplished
© Copyright 2016 OSIsoft, LLCUSERS CONFERENCE 2016
Contact Information
Kyle Gogolinski
Team Leader, Process Control & Automation: Systems (Upgrading / Utilities)
Syncrude Canada Ltd.
27
Peter Wright
Manager, Industrial Information Practice
Dexcent Inc.
27
© Copyright 2016 OSIsoft, LLCUSERS CONFERENCE 2016
Questions
Please wait for the
microphone before asking
your questions
Please remember to…
Complete the Online Survey
for this session
State your
name & company
28
http://ddut.ch/osisoft
search OSISOFT in the app store
© Copyright 2016 OSIsoft, LLCUSERS CONFERENCE 2016
Thank You
© Copyright 2016 OSIsoft, LLCUSERS CONFERENCE 2016
